Multimodal Understanding on RWQA (RWQA score and Rel. Avg.)
60.2RWQA ScoreVanilla
Evaluation results
| Method | Links | ||
|---|---|---|---|
| VanillaModels=BAGEL-7B, Avg. Tokens=100%2026.05 | 60.2 | 100 | |
| G2TRModels=BAGEL-7B, Avg. Tokens=50%2026.05 | 59 | 99 | |
| IVC-PruneModels=BAGEL-7B, Avg. Tokens=50%2026.05 | 57.8 | 98.6 | |
| W-FastVModels=BAGEL-7B, Avg. Tokens=54%2026.05 | 55.4 | 95.6 | |
| FastVModels=BAGEL-7B, Avg. Tokens=54%2026.05 | 55.2 | 95.8 | |
| PDropModels=BAGEL-7B, Avg. Tokens=61%2026.05 | 53.8 | 94.1 | |
| VanillaModels=InternVL-U, Avg. Tokens=100%2026.05 | 48.8 | 100 | |
| VSCANModels=BAGEL-7B, Avg. Tokens=50%2026.05 | 48.4 | 94.4 | |
| G2TRModels=InternVL-U, Avg. Tokens=50%2026.05 | 46.8 | 94.9 | |
| IVC-PruneModels=InternVL-U, Avg. Tokens=50%2026.05 | 46 | 93.7 | |
| PDropModels=InternVL-U, Avg. Tokens=61%2026.05 | 45.6 | 92 | |
| FastVModels=InternVL-U, Avg. Tokens=54%2026.05 | 45.5 | 92.9 | |
| W-FastVModels=InternVL-U, Avg. Tokens=54%2026.05 | 45.2 | 92.3 | |
| VSCANModels=InternVL-U, Avg. Tokens=50%2026.05 | 43.5 | 92.2 |